Hybrid Uncertainties Quantification And Identification Method Based On Power Spectral Density Response

In order to improve the prediction accuracy,the dynamics simulation of complex structures must consider many uncertain factors,such as the non-uniformity of structural materials,the instability of working environment,and the model deviation caused by simplified assumptions in modeling and simulation.How to identify and quantify the uncertainty of different sources in modeling and simulation is one of the core problems of model validation.Uncertainty model modification is developed on the basis of traditional structural dynamics model modification,which has certain guiding significance on how to identify and quantify the uncertainty from various sources.However,the existing uncertainty model correction framework mostly discusses the problem that the response quantity is a scalar,which is not applicable to dynamic response problems such as power spectral density function.This paper reviews the uncertainty identification and quantization framework in scalar form,and improves the uncertainty quantization and identification framework by using the probabilistic box method as the quantization tool to make it suitable for the uncertainty quantization and identification of power spectral density response problems.Specific research is carried out in the following aspects:1.The mixed uncertainty quantization of numerical value,parameter and model form is studied by probability box method.The overall framework and process of the probabilistic box method for multi-source uncertainty quantification were defined,and the multi-source uncertainty was quantified for the 2014 sandia model validation and validation challenge.The research shows that the probabilistic box method can effectively manage and quantify the multi-source uncertainty problems with scalar quantities.2.Based on the probabilistic box method,a multi-response model validation metric is proposed.Firstly,the probability of mahalanobis distance is introduced to reduce the dimensionality of the multidimensional response quantity,and the multidimensional validation measurement problem is transformed into the comprehensive measurement problem of mahalanobis distance by taking mahalanobis distance as the conversion statistic,and the method is verified by numerical examples and 2014 sandia model validation and validation challenge problem.The results show that the multi-response validating metric based on probability box and markov distance is correct and stable.3.Summarized the validation measure:ment methods of power spectral density response under the deterministic framework,and pointed out the applicable scope of various methods;An area measurement method for power spectral density response under uncertainty frame is proposed.In view of the deterministic model validation measurement method,based on the correlation coefficient method widely used in structural dynamics model modification,the concept of taking correlation coefficient as the indicator as the confirmation measurement is further clarified,and the applicable scope of various correlation coefficients as the confirmation measurement is summarized.The probabilistic box area measurement method is extended to confirm the power spectral density response.4.In order to achieve the separation of parameters and model form uncertainties,an improved multi-source uncertainty quantization and identification method was proposed based on the power spectral density response.The solution flow of the improved multi-source uncertainty quantization and identification method is given,and the uncertainty quantization and identification of the power spectral density response of the frame structure with multiple bolted connections is carried out,and the power spectral density response under different excitations is predicted,and the proposed method is confirmed.